1. Field of the Invention
This invention relates to a speech encoding method and apparatus in which an input speech signal is divided in terms of blocks or frames as encoding units and encoded in terms of the encoding units, and an audio signal encoding method and apparatus in which an input audio signal is encoded by being represented with parameters derived from a signal corresponding to an input audio signal converted into a frequency range signal.
2. Description of the Related Art
There have hitherto been known a variety of encoding methods for encoding an audio signal (inclusive of speech and acoustic signals) for signal compression by exploiting statistic properties of the signals in the time domain and in the frequency domain and psycho acoustic characteristics of the human being. The encoding method may roughly be classified into time-domain encoding, frequency domain encoding and analysis/synthesis encoding.
Examples of the high-efficiency encoding of speech signals include sinusoidal analytic encoding, such as harmonic encoding or multi-band excitation (MBE) encoding, sub-band coding (SBC), linear predictive coding (LPC), discrete cosine transform (DCT), modified DCT (MDCT) and fast Fourier transform (FFT).
Meanwhile, in representing an input audio signal, such as speech or music signals, with parameters derived from a signal corresponding to the audio signal transformed into a frequency range signal, the commonplace practice is to quantize the parameters by weighted vector quantization. These parameters include frequency range parameters of the input audio signal, such as discrete Fourier transform (DFT) coefficients, DCT coefficients or MDCT coefficients, amplitudes of harmonics derived from these parameters and harmonics of LPC residuals.
In carrying out weighted vector quantization of these parameters, the conventional practice has been to calculate frequency characteristics of the LPC synthesis filter and that of the perceptually weighting filter to multiply them by each other or to calculate the frequency characteristics of the numerator and the denominator of the product to find a ratio thereof.
However, in calculating the weight value for vector quantization, a large number of processing operations are generally involved, such that it has been desired to reduce the processing volume further.
It is therefore an object of the present invention to provide speech encoding method and apparatus and an audio signal encoding method and apparatus for reducing the processing volume involved in calculating the weight value for vector quantization.
According to the present invention, there is provided a speech encoding method in which an input speech signal is divided on the time axis in terms of pre-set encoding units and encoded in terms of the pre-set encoding units. The method includes the steps of finding short-term prediction residuals of the input speech signal, encoding the short-term prediction residuals thus found by sinusoidal analytic encoding and encoding the input speech signal by waveform encoding. The perceptually weighted vector quantization or matrix quantization is applied to sinusoidal analysis encoding parameters of the short-term prediction residuals and, at the time of the perceptually weighted vector quantization or matrix quantization, the weight value is calculated based on the results of an orthogonal transform of parameters derived from the impulse response of the transfer function of the weight value.
With the method for encoding an audio signal in which an input audio signal is represented with parameters derived from a signal corresponding to the input audio signal transformed into a frequency range, the weight value for weighted vector quantization of the parameters is calculated based on the results of orthogonal transform of parameters derived from the impulse response of the transfer function of the weight.